OpenAI Rolls Out ChatGPT Images 2.0 with Thinking Capabilities

Gate News message, April 22 — OpenAI began rolling out ChatGPT Images 2.0 on April 21 to all ChatGPT and Codex users, with paid users able to generate more advanced outputs. The company launched the gpt-image-2 API with pricing based on output quality and resolution.

The new model improves rendering of small text and UI elements, supports scripts including Japanese, Korean, Hindi, and Bengali, and follows detailed instructions more precisely. Images 2.0 introduces “thinking” capabilities, allowing the model to search the web for real-time information, generate multiple distinct images from a single prompt, and refine outputs for accuracy and consistency. Complex generations can take several minutes; the model has a knowledge cutoff of December 2025.

OpenAI did not disclose the model’s architecture. For pricing context, Google’s Imagen 4 costs $0.02 to $0.06 per image, while generating 10,000 images with GPT Image 1 Mini in low-quality mode costs approximately $50, compared to about $1,670 for high-quality output.